Translation and Localisation Checklist: Grant Scope, Vendor Ownership, Moral Rights, Marks and Transliteration, and Territory Control

By Casey Scott McKay · 2024-10-09

This checklist builds a localisation position that produces owned assets instead of a decade of unattributed derivative works. Phase one establishes the authority to translate before anything is commissioned, because a translation made without it infringes and cannot be protected. Phase two fixes ownership using the work made for hire category that applies to translations and almost nothing else, with an assignment in the alternative and flow-down to the freelancers who actually do the work. The middle phases separate the audiovisual rights bundle, map moral rights across target markets, and draft language and territory as distinct grants. The closing phases cover brand localisation, software strings, machine translation and synthetic voice policy, regulated content review, registration, and the chain of title record a buyer will sample.

IP and Technology > Copyright | Checklist | Published 9 October 2024 - Updated 26 February 2026 | Casey Scott McKay - marksy.us

Summary. This checklist builds a localisation position that produces owned assets instead of a decade of unattributed derivative works. Phase one establishes the authority to translate before anything is commissioned, because a translation made without it infringes and cannot be protected. Phase two fixes ownership using the work made for hire category that applies to translations and almost nothing else, with an assignment in the alternative and flow-down to the freelancers who actually do the work. The middle phases separate the audiovisual rights bundle, map moral rights across target markets, and draft language and territory as distinct grants. The closing phases cover brand localisation, software strings, machine translation and synthetic voice policy, regulated content review, registration, and the chain of title record a buyer will sample.

Keywords: localisation checklist · authority to translate · work made for hire writing · present tense assignment · subcontractor flow down · translation memory ownership · dubbing rights bundle · performer reuse terms · subtitle source files · moral rights mapping · adaptation permission grading · language and territory grants · automatic reversion · brand transliteration · foreign equivalents · machine translation policy · synthetic voice consent · regulated content review · translation registration · chain of title record

How to use this checklist

| Field | Detail | |---|---| | Who runs it | IP counsel with localisation, procurement, product, and marketing | | When | Before a market launch; on vendor selection or renewal; annually across the estate | | Time required | Two weeks for the forward position; longer for historic remediation | | Gates | Authority recorded before commissioning; ownership terms executed before delivery | | Output | A rights register, compliant vendor terms, a machine translation policy, and a chain of title record | | Companion documents | Localising Content for Foreign Markets and Saying It in Another Language |

The matter. A software business with eleven years of history localises its product, documentation, help content, marketing site, and training videos into fourteen languages. Localisation is bought by procurement on purchase orders that describe a per-word rate and nothing else. Three vendors have been used over that period, each subcontracting to freelancers in market. The current vendor holds a translation memory containing the company's entire terminology history and has begun using a machine translation engine, sending client content to a third-party service. Marketing has adopted an informal local brand name in one market that nobody has filed. The training videos are dubbed, and the company wants to reuse the voice tracks in a new product launch. A buyer has begun diligence.

Phase 1. Establish the authority to translate

[ ] Confirm the right for each underlying work. Ownership, a licence expressly including translation and adaptation, or a statutory exception. Why. Translating a protected work is preparing a derivative work under 17 U.S.C. § 106(2), and 17 U.S.C. § 101 lists translation first among the examples. Trap. Assuming that owning a copy or holding a distribution licence permits translation.
[ ] Read for adaptation as well as translation. Why. A right to translate is not a right to adapt, and localisation almost always adapts. Trap. A licence granting "translation rights" relied on for a culturally reworked version.
[ ] Check whether the original is in the public domain. Why. It may be translated freely and the translation is a fully protected new work. Trap. Paying for a licence that is unnecessary.
[ ] Treat orphan originals as an orphan works problem. Diligent search, documented, with a risk tier. Why. There is no mechanism to authorise the translation. Trap. Proceeding because the rights holder cannot object if it cannot be found.
[ ] Note the double disadvantage of an unauthorised translation. Why. It infringes, and 17 U.S.C. § 103(a) denies protection to the part using the preexisting material unlawfully. Trap. Building a business on a translation that cannot be enforced.
[ ] [Gate] Record the authority in the project file before commissioning. Trap. Commissioning first and clearing later, which is the sequence in most organisations.

Phase 2. Fix ownership

[ ] Use the work made for hire writing. Why. 17 U.S.C. § 101 lists translation among the nine categories in which a specially commissioned work can be a work made for hire, with a written instrument signed by both parties expressly agreeing it. Trap. A purchase order, which is the single most common cause of lost ownership in this field.
[ ] Add a present-tense assignment in the alternative. "Hereby assigns", covering translations, subtitle files, recordings, glossaries, and derivative material. Why. The category test and the writing requirement both fail more often than expected. Trap. "Agrees to assign", which leaves title with the assignor.
[ ] Comply with 17 U.S.C. § 204. A transfer must be in writing and signed by the owner. Trap. An assignment in a vendor's master agreement that does not reach the individual who did the work.
[ ] Understand the termination difference. Why. Works made for hire are outside 17 U.S.C. § 203; assigned works are not, so an assignment can be terminated decades later. Trap. Treating the two routes as equivalent.
[ ] Flow down to subcontractors with evidence. Why. Vendors subcontract to freelancers in market routinely, and this is where chains break. Trap. A flow-down obligation with no right to see the signed documents.
[ ] Waive moral rights to the extent permitted, recognising that several jurisdictions treat them as inalienable. Trap. A global waiver assumed to be effective everywhere.
[ ] Require native-format deliverables. Editable subtitle files, unmixed dub stems, resource files rather than compiled builds. Why. A rendered deliverable cannot be re-versioned. Trap. Accepting burned-in subtitles.
[ ] [Gate] No delivery accepted before the ownership terms are executed. Trap. Work beginning while the contract is "in legal".

Phase 3. Separate the audiovisual bundle

[ ] Treat the script translation, the dub recording, the performer contributions, the music synchronisation, and the subtitle file as five things. Why. They are separately created, separately owned, and separately licensed. Trap. One purchase order covering "localisation of the video".
[ ] Record performer terms per language. Buyout scope, reuse rights across media and territories, residual obligations. Why. Reuse of a localised voice line in a trailer, a sequel, or a new product is a scope question nobody asked at recording. Trap. Assuming a buyout covers everything.
[ ] Re-check music synchronisation where the localised version alters the audio bed. Trap. A cleared sync licence relied on for a re-mixed dub.
[ ] Own the subtitle files. Why. They are literary works whose timing and condensation involve judgment. Trap. Subtitles produced by a post house under a production services agreement silent on copyright.
[ ] Handle community translation through the terms of service, with an assignment contributors have actually accepted and the acceptance recorded. Trap. Volunteer subtitles adopted into a commercial release.

Phase 4. Map moral rights by market

[ ] Assume moral rights everywhere except at home. Why. The United States gives them only to works of visual art under 17 U.S.C. § 106A; most markets give them to authors generally. Trap. A domestic analysis applied globally.
[ ] Identify the integrity risk in each target market. Why. Heavy adaptation, cultural substitution, censorship edits, and poor translation can each be prejudicial modification. Trap. Treating adaptation as a creative decision with no legal dimension.
[ ] Preserve attribution. Why. The attribution right requires the author to be named, and localised editions routinely omit or obscure it. Trap. A credit lost in a layout change.
[ ] Check whether waiver is available in each market rather than assuming the contract settles it. Trap. A waiver governed by one law offered as an answer under another.
[ ] Remember the translator's moral rights. Why. An in-house edit to a delivered translation can generate a claim from the translator in their home jurisdiction. Trap. Editorial revision treated as internal work.
[ ] Use Itar-Tass Russian News Agency v. Russian Kurier to frame ownership questions and Subafilms v. MGM-Pathe Communications for the extraterritoriality limit. Trap. Assuming domestic law governs a foreign-market dispute.
[ ] Grade the adaptation permission in every licence. Translation only; translation with minor adjustment; adaptation with approval; free adaptation. Trap. Silence, which each party reads in its own favour.

Phase 5. Draft language and territory separately

[ ] Grant language and territory as distinct terms. Why. Spanish, Portuguese, French, and Arabic each span multiple markets with material differences. Trap. A bare language grant covering every market where the language is spoken.
[ ] Address digital availability expressly. Why. A digital edition sold in one market is reachable from every other, and the technology does not respect the grant. Trap. Territorial exclusivity granted twice for the same digital edition.
[ ] Say what happens to adaptations of a translation for a neighbouring market, and who owns them. Trap. European Spanish reworked for Mexico with no allocation.
[ ] Include automatic reversion on non-exercise or lapse, without notice. Why. Grantees do not volunteer reversion. Trap. A reversion clause requiring the grantor to monitor exploitation.
[ ] Address sublicensing and proceeds split. Trap. A sublicence granted into a medium the head grant never covered.

Phase 6. Localise the brand

[ ] Choose the strategy deliberately. Keep, transliterate, translate, or coin. Trap. Letting a distributor decide.
[ ] Expect the doctrine of foreign equivalents. Why. A term descriptive in another language may be refused under 15 U.S.C. § 1052(e). Trap. Filing a translated mark without a descriptiveness assessment.
[ ] Treat transliteration as a new mark with its own meaning, phonetics, and registrability. Trap. A phonetic rendering that carries an unintended meaning.
[ ] File the local form before someone else does. Why. A market that coins an informal local name will use it, and a third party will register it. Trap. An informal name used for years and owned by nobody.
[ ] Clear slogans separately, since they frequently become descriptive or laudatory in translation. Trap. A tagline registered at home and unprotectable abroad.
[ ] Run native-speaker review that is not the translator. Trap. Negative associations discovered at launch.
[ ] Secure local-language domains and handles with the mark. Trap. A local distributor holding the domain.

Phase 7. Software, strings, and assets

[ ] Treat interface strings as literary works and their translations as derivative works. Trap. String files excluded from the ownership clause because they are "code".
[ ] Account for layout consequences. Right-to-left scripts, character sets, text expansion, locale formats. Why. Where the interface carries protectable visual expression, the localised version is a derivative of it. Trap. Layout work commissioned separately with no ownership terms.
[ ] Check open source obligations. Why. Copyleft licences reach translated string files forming part of the licensed work. Trap. Translations of a copyleft component treated as proprietary.
[ ] Negotiate translation memory and glossary ownership, use restrictions, and delivery format on termination. Why. It is the most valuable asset in a long-running relationship and is usually conceded by silence. Trap. A vendor change blocked by a memory the vendor owns.
[ ] Address model training expressly. Whether the vendor may train on client content and who owns the result. Trap. A clause covering confidentiality but not training.
[ ] Map content types to vendors. Product, documentation, help, store listings, marketing, community. Why. Different vendors fragment ownership across one product experience. Trap. An audit that covers the product and misses the store listing.
[ ] Record chain of title per asset in the localisation system, not in the contract file. Trap. A record that cannot be queried by asset.

Phase 8. Machine translation and synthetic voice

[ ] Write the machine translation policy down. Why. In its absence the default is unrestricted use. Trap. A prohibition assumed and never communicated.
[ ] Recognise that raw output likely has no author and that post-editing protects only the human contribution under 17 U.S.C. § 103(b). Trap. Relying on translation copyright to enforce against copying of machine-translated content.
[ ] Note that the underlying rights are unaffected. Why. Machine-translating a protected work is still preparing a derivative work. Trap. Treating automation as authority.
[ ] Prohibit submission of confidential content to public services and require disclosure of engines, processing locations, and retention. Trap. Unpublished material passing through a consumer service.
[ ] Set quality gates by content class. Light review for support content; qualified in-market review for safety, contractual, and regulated content. Trap. One quality standard across all content.
[ ] Carve regulated and safety-critical content out of the liability cap. Why. Standard terms cap at the fee, which is not a risk transfer where a mistranslated warning causes injury. Trap. A cap accepted without reference to content class.
[ ] Build synthetic voice consent architecture. Which voice, which content, which languages, how long, further derivatives, and survival after death. Why. Digital replica statutes create express rights and existing performer contracts rarely address it. Trap. Training a model on archived recordings under an old buyout.
[ ] Check collective agreements where entertainment sector arrangements addressing synthetic performance apply. Trap. Obligations assumed to bind only the producer.

Phase 9. Regulated content, registration, and records

[ ] Identify content with statutory translation obligations. Labelling, instructions, warnings, contract terms, employment and privacy documents. Trap. Compliance translation bought on the same terms as marketing copy.
[ ] Require in-market qualified review for regulated content. Why. The requirement is regulatory accuracy rather than linguistic fidelity. Trap. A linguistically perfect translation that is regulatorily wrong.
[ ] Resolve the prevailing version question where a contract exists in two languages and local law requires a local version. Trap. A prevailing-version clause that local law will not give effect to.
[ ] Re-examine advertising claims per market. Comparative rules, superlative restrictions, and prohibited terms differ. Trap. A substantiated claim translated into an unsubstantiated one.
[ ] Register the translations that matter. Why. 17 U.S.C. § 411 makes registration a precondition to suit and 17 U.S.C. § 504 statutory damages depend on timely registration of the translation itself. Trap. Registering only the original.
[ ] Keep a per-asset chain of title record. Language, vendor, contract reference, ownership basis, delivery date, subcontractor evidence, registration status. Why. A buyer will sample it. Trap. A record built during diligence.

Phase 10. Working the example matter

[ ] Fix the forward position this week. Why. Eleven years of purchase orders is a large remediation project, and every new purchase order makes it larger. Adding ownership, subcontracting, confidentiality, memory, and deliverable clauses to the standard engagement terms stops the growth immediately. Trap. Beginning with the historic audit, which takes months while the estate keeps expanding.
[ ] Deal with the current vendor's machine translation practice at once. Why. Client content — product strings, documentation, potentially unreleased material — is being sent to a third-party service under terms nobody has read. That is a confidentiality exposure independent of any copyright question. Trap. Treating it as a quality issue.
[ ] Secure the translation memory before renewal. Why. Eleven years of terminology sits with a vendor, and the company's ability to change vendors depends on getting it. Renewal is the only moment of leverage. Trap. Renewing on price and raising ownership afterwards.
[ ] Sample the chain of title across ten high-value assets. Product strings, core documentation, and the training videos. Why. The failure rate predicts the population and gives the buyer a quantified answer rather than an admission. Trap. Attempting a complete audit under diligence time pressure.
[ ] Run confirmatory assignments through the vendors still engaged. Why. They will cooperate while the relationship continues; the two former vendors will be harder and their freelancers harder still. Trap. Starting with the oldest and least reachable population.
[ ] Stop the dub reuse until the performer terms are checked. Why. Reusing localised voice tracks in a new product launch is a scope question, and buyout language from years ago may not reach a different product. Trap. Reuse authorised by the localisation team on the basis that the tracks were paid for.
[ ] File the informal local brand name now. Why. A name the market has adopted and nobody has registered is available to a third party, and it is the name customers use. Trap. Filing the official brand name and leaving the one in actual use unprotected.
[ ] Give the buyer a remediation plan rather than an assurance. Scope, completion percentage, quantified residual, and the date the forward position was fixed. Why. It is materially more reassuring than a claim that everything is in order, and it is verifiable. Trap. Warranting a chain of title the company has not tested.
[ ] [Gate] Sequence: forward terms, confidentiality, memory, sample, assignments, dub reuse, brand filing. Trap. Sequencing by what diligence asked about first.

Phase 11. Publishing and content licensing

[ ] Verify the head grant, not the sublicensor's summary. Why. A foreign publisher's rights are a sublicence and can be no broader than what the original publisher held. Trap. A summary that omits a media carve-out.
[ ] Note that the translator contracts separately. Why. A publisher can hold the right to publish a translation and not own the translation it paid for. Trap. Assuming the publishing licence resolves the translator's position.
[ ] Address translator credit and royalty participation as a commercial term rather than a courtesy. Trap. A flat fee assumed to be the market position when it is being renegotiated across the trade.
[ ] Build automatic reversion on non-exercise or out-of-print status. Trap. Reversion requiring notice from a grantor with no visibility of exploitation.
[ ] Separate audio, dramatic, and film rights in the translation from the right to publish it. Trap. A silent translation licence leaving them with the head grantor and blocking an adaptation.
[ ] Clear both copyrights for quoted extracts. Why. A translated passage engages the translation and the original. Trap. A permissions process that clears one.
[ ] Remember that a public domain original does not simplify the licence chain. Why. A new translation of a public domain work is fully protected and separately owned. Trap. Assuming free use of an existing translation of a classic.

Phase 12. Games, apps, and continuous content

[ ] Plan for volume. Hundreds of thousands of words across dialogue, interface, marketing, store listings, patch notes, and community content, updated continuously, in many languages. Why. The ownership problem scales with the volume. Trap. A process designed for a single manual.
[ ] Expect vendor fragmentation across years. Why. Different vendors, different contracts, some with assignments and some without, all inside one build. Trap. An audit that samples the current vendor only.
[ ] Record performer terms per language. Why. Each localised recording carries its own contributions and its own reuse scope. Trap. A trailer using a localised line recorded for in-game dialogue.
[ ] Assign community translation through accepted terms of service and record the acceptance. Trap. Volunteer work adopted into a shipped build with no assignment.
[ ] Treat culturalisation as adaptation. Why. Changes for local sensibilities, regulation, or age rating create further derivative works and can raise integrity claims. Trap. Classification edits made under a licence prohibiting modification.
[ ] Include store listings and metadata. Why. They are separate works, frequently produced by a marketing vendor under a contract that never mentions copyright. Trap. A rights audit stopping at the product.
[ ] Put chain of title into the localisation pipeline. Per string: vendor, contract, ownership basis. Why. It is the only thing that answers the question in year six. Trap. A contract repository with no per-asset link.

Phase 13. The vendor agreement, clause by clause

[ ] Ownership. Work made for hire where the category applies, present-tense assignment in the alternative, covering translations, subtitle files, recordings, glossaries, and derivative material, with further assurances and a moral rights waiver to the extent permitted. Trap. An ownership clause covering "deliverables" without defining them.
[ ] Subcontracting. Consent or notification, equivalent flow-down, and evidence of signed assignments on request. Trap. Flow-down with no verification right.
[ ] Confidentiality and data handling. Third-party service use, processing locations, retention limits, deletion on termination. Trap. A confidentiality clause silent on machine translation engines.
[ ] Translation memory and glossary. Ownership, permitted use for other clients, delivery on termination, and format. Trap. Delivery promised in a proprietary format that is useless elsewhere.
[ ] Quality and liability. Accuracy standard, review process, and a carve-out from the fee cap for safety-critical and regulated content. Trap. A single cap across all content classes.
[ ] Deliverables. Source-form files rather than rendered output. Trap. Compiled builds and burned-in subtitles.
[ ] Performer terms for audiovisual work: buyout scope, reuse, residuals, and any synthetic voice permission with its own consent scope. Trap. Performer terms handled by a studio the client never contracts with.
[ ] Termination and transition. Return or deletion, delivery of memories and glossaries, transition assistance at agreed rates. Trap. No transition obligation, in a market where vendor changes are frequent.

Phase 14. Governance

[ ] Name an owner across procurement, product, marketing, and legal. Why. Localisation is commissioned by all four and owned by none. Trap. Ownership assigned to the function that spends the most.
[ ] Keep three registers. Rights to translate each underlying work; vendors and their terms; localised assets with ownership basis and registration status. Trap. Registers that cannot be joined.
[ ] Gate two moments. Before commissioning a new work for localisation, confirm the authority. Before engaging a vendor, confirm the terms. Trap. Gates applied to large projects only.
[ ] Report three measures. Assets with a recorded ownership basis; vendors with compliant terms including flow-down; open confirmatory assignments. Trap. Reporting words translated.
[ ] Review annually. New markets, new vendors, machine translation practice against policy, and content that has moved into a regulated category. Trap. A review that checks the policy exists.

Phase 15. Inheriting an existing estate

[ ] Triage by asset life. Product documentation, interface strings, catalogue titles, and training material justify remediation; campaign copy from three years ago does not. Trap. Uniform remediation across everything.
[ ] Fix the forward position before the backward one. Why. New terms applying to all future work stop the problem growing while historic remediation runs. Trap. An audit that runs for six months while purchase orders continue.
[ ] Work backwards through vendors still engaged, then their freelancers. Why. Cooperation follows the commercial relationship. Trap. Starting with the oldest and least reachable population.
[ ] Accept an untraceable residual. Why. A freelancer engaged through an agency eight years ago may be unreachable, and the risk must be quantified rather than pursued indefinitely. Trap. A remediation programme with no completion criterion.
[ ] Assess practical exposure alongside theoretical exposure. Why. A paid translator with no ongoing relationship whose work is embedded in a larger product is unlikely to assert. Trap. Treating practical assessment as a substitute for the forward fix.
[ ] Fix registration for the assets that will be enforced. Trap. Registering everything, which is neither necessary nor affordable.
[ ] Document the programme. Scope, completion percentage, quantified residual. Why. A buyer is more reassured by a measured programme than by an untested assurance. Trap. A warranty offered in place of evidence.

Phase 16. Accessibility alongside localisation

[ ] Distinguish captions from subtitles. Why. Captions for accessibility may fall within statutory exceptions in some circumstances; subtitles for localisation are ordinary derivative works requiring authority. Trap. One production process treated as one legal category.
[ ] Treat audio description as a new work. Written, performed, recorded, with its own author, performer, and ownership question. Trap. Description commissioned with no copyright terms because it is "accessibility work".
[ ] Apply exception conditions strictly where relied on. Authorised entity status, eligible beneficiaries, format limits, record-keeping. Why. A commercial localisation programme rarely qualifies. Trap. An exception invoked for a commercial release.
[ ] Handle sign language interpretation as performance and translation at once. Trap. Interpreter contributions omitted from the rights bundle.
[ ] Draft the ownership terms once to cover all four outputs. Captions, subtitles, description, and dubbed tracks. Trap. Four inconsistent contracts for one production process.
[ ] Keep the accessibility record separate. Why. Where an exception is relied on, the conditions and eligibility evidence must be documented, and mixing it into a commercial file makes it unusable. Trap. One combined project file.

Phase 17. Enforcement

[ ] Classify the dispute before choosing a remedy. Unauthorised translation, ownership gap, or scope breach. Why. They have entirely different remedies and only the first is a straightforward infringement claim. Trap. Treating an ownership gap as something a demand letter can fix.
[ ] For an unauthorised translation, claim under 17 U.S.C. § 106(2) and note that 17 U.S.C. § 103(a) denies the infringer protection for the infringing portion. Trap. Overlooking that the infringer's translation can be used by the rights holder.
[ ] For an ownership gap, expect negotiation rather than remedy. Why. The translator holds a right the commissioner needs and the commissioner has already exploited the work. Trap. Litigating from the weaker position.
[ ] For a scope breach, plead contract first and infringement where the breach exceeds the licence. Trap. An infringement claim where the licence in fact covered the use.
[ ] Register before suing. 17 U.S.C. § 411 is a precondition and 17 U.S.C. § 504 statutory damages depend on timely registration of the translation. Trap. Registration of the original only.
[ ] Litigate foreign exploitation abroad. Why. Subafilms v. MGM-Pathe Communications confirms the Act does not reach acts occurring wholly abroad. Trap. A domestic claim about a foreign-market edition.
[ ] Model the ownership law of the relevant forum. Why. Itar-Tass Russian News Agency v. Russian Kurier applies the law of the country with the most significant relationship, which can produce a different owner. Trap. Assuming the contract's governing law settles ownership.
[ ] Use platform takedown for pirated localised editions, which requires evidencing ownership of the translation rather than only the original. Trap. A takedown refused for want of proof of the translation's ownership.

Phase 18. Documents this checklist should produce

[ ] A rights register recording, per underlying work, the authority to translate and its source. Why. It is the precondition for everything else. Trap. A register listing works without the authority reference.
[ ] Standard engagement terms containing ownership, subcontracting, confidentiality, memory, quality, deliverable, performer, and transition clauses. Why. One document fixes the forward position permanently. Trap. Terms drafted and not applied to purchase orders.
[ ] A vendor register recording current terms, flow-down status, memory ownership, and machine translation practice. Trap. A supplier list with no contract detail.
[ ] A moral rights map for the principal markets, recording whether waiver is available and what adaptation permission has been graded. Trap. A single global answer.
[ ] A machine translation and synthetic voice policy, written, communicated, and reflected in vendor terms. Trap. A policy that exists only in legal's files.
[ ] A per-asset chain of title record inside the localisation system, queryable by asset, language, and vendor. Why. A buyer will sample it. Trap. A record assembled during diligence.
[ ] A registration schedule for the translations that will be enforced. Trap. Registration decided asset by asset with no policy.

Phase 19. Proportion and sequencing

[ ] Do the forward fix first, always. Why. It costs a paragraph and stops the estate growing while everything else is sorted. Trap. Beginning with the historic audit because it is what diligence asked about.
[ ] Sort content by expected life. Long-lived content justifies the full architecture; short-lived content does not. Why. One process applied to both is abandoned within a year. Trap. A uniform standard adopted for consistency.
[ ] Do four things regardless of scale. Record the authority, use an ownership clause, require source-form deliverables, and keep a per-asset record. Why. Together they take an afternoon to establish and prevent most of what goes wrong. Trap. Skipping them for small projects, which is how an estate accumulates thousands of unattributed assets.
[ ] Assign owners and dates to every open item. Why. This checklist generates enough findings to disappear without a tracker. Trap. A report circulated for information.

Phase 20. Adaptation beyond translation

[ ] Identify transcreation as authorship. Why. Rewriting a message to achieve the same effect in another culture produces a new work rather than a translation of anything. Trap. Transcreation bought on translation terms and owned by the copywriter.
[ ] License format adaptation separately. Why. Remaking a series, book, or game for a local market is a derivative work of the underlying property, with the local production's contributions owned locally. Trap. Format rights assumed to travel with translation rights.
[ ] Assess cultural substitution against integrity rights. Why. Replacing references, names, settings, and humour can be extensive enough to be prejudicial modification. Trap. Substitution decided by a local marketing team with no rights review.
[ ] Address compliance and classification edits in the licence. Why. Modifications required by local law or platform policy can conflict with a term prohibiting modification. Trap. A conflict discovered at release.
[ ] Anticipate reverse localisation. Why. Content made for one market and adapted for a global audience uses the same analysis, and the commissioning bundle rarely contemplated it. Trap. A global release blocked by a domestic commissioning agreement.
[ ] Grade the permission by degree in every grant. Translation only; minor adjustment; adaptation with approval; free adaptation. Trap. A single word — "localise" — doing the work of four distinct permissions.

Phase 21. Questions the business will ask

[ ] "We paid for it, so we own it." Why the answer is no. 17 U.S.C. § 204 requires a signed writing for a transfer, and payment alone transfers nothing for a commissioned translation. Trap. An answer that stops there rather than explaining the work made for hire route that is available.
[ ] "Can we just machine translate it?" Why the answer is nuanced. The output may have no author, but translating a protected work is still preparing a derivative work under 17 U.S.C. § 106(2). Trap. Conflating the protection question with the authority question.
[ ] "The author signed a waiver." Why to check where. Several jurisdictions treat moral rights as inalienable. Trap. A waiver relied on across every market.
[ ] "We have Spanish rights." Why to ask which territories. A bare language grant covers every market where the language is spoken. Trap. A second licensee sold rights the first already holds.
[ ] "Can we reuse the dub?" Why it is three questions. Performer terms, music synchronisation, and the underlying grant. Trap. One answer covering all three.
[ ] "Do we need to register the translations?" Why yes, for anything enforced. 17 U.S.C. § 411 and 17 U.S.C. § 504. Trap. Registration treated as optional because the original is registered.

Outcome. A rights register recording the authority to translate each work; engagement terms that make every new translation an owned asset, with flow-down evidenced; the audiovisual bundle separated and performer reuse recorded; a moral rights map for the principal markets; language and territory granted distinctly with automatic reversion; a written machine translation and synthetic voice policy; and a per-asset chain of title record that survives a diligence sample.

The five things people get wrong

One. Buying localisation on a purchase order. It is the single largest cause of lost ownership in this field. Translation is one of the nine work made for hire categories — an advantage available almost nowhere else in copyright practice — and it is squandered by an engagement document that mentions only a per-word rate.

Two. Stopping the paper at the vendor. Vendors subcontract to individual freelancers in the target market as a matter of routine. Terms that bind the vendor and do not reach the person who actually made the expressive choices deliver nothing, and the gap is invisible until someone asks for the signed assignment.

Three. Treating localisation as one thing. A dubbed video involves a script translation, a sound recording, performer contributions, music synchronisation, and a subtitle file. A single line item in a budget conceals five separate rights, and the one that blocks reuse is usually the performer terms.

Four. Assuming the moral rights position travels. The domestic position is exceptional. In most target markets authors hold integrity and attribution rights, waiver may be unavailable, and a heavily adapted edition made under a licence governed by a permissive law can still be unlawful where it is published.

Five. Letting the vendor own the translation memory. A decade of a company's terminology, style, and prior decisions sits in a file the vendor holds, and it is discovered at the moment the company wants to change vendors. It is a single clause and it is almost never negotiated.

Key Authorities at a Glance

| Authority | Proposition | |---|---| | 17 U.S.C. § 101 | Derivative work; work made for hire categories | | 17 U.S.C. § 102 | Subject matter; originality | | 17 U.S.C. § 103 | Derivative works; scope of protection | | 17 U.S.C. § 106 | Exclusive rights | | 17 U.S.C. § 106A | Moral rights in visual art | | 17 U.S.C. § 107 | Fair use | | 17 U.S.C. § 201 | Ownership | | 17 U.S.C. § 203 | Termination of transfers | | 17 U.S.C. § 204 | Writing requirement | | 17 U.S.C. § 411 | Registration precondition | | 17 U.S.C. § 504 | Statutory damages | | 15 U.S.C. § 1052 | Refusals; foreign equivalents | | 15 U.S.C. § 1125 | False designation | | Feist Publications v. Rural Telephone Service | Originality | | Community for Creative Non-Violence v. Reid | Employee status | | Stewart v. Abend | Derivative works and underlying rights | | Google LLC v. Oracle America | Fair use and purpose | | Andy Warhol Foundation v. Goldsmith | Shared purpose | | Gracen v. Bradford Exchange | Derivative originality | | Durham Industries v. Tomy | Distinguishable variation | | Woods v. Bourne | Derivative protection scope | | Itar-Tass Russian News Agency v. Russian Kurier | Choice of law for ownership | | Subafilms v. MGM-Pathe Communications | No extraterritorial application | | Berne translation right | International translation right | | Moral rights abroad | Integrity and attribution | | Digital replica statutes | Synthetic voice | | Translation memory ownership | Localisation assets | | Regulated labelling translation | Compliance translation |